https://sourceware.org/bugzilla/show_bug.cgi?id=7023
--- Comment #14 from Sourceware Commits <cvs-commit at gcc dot gnu.org> --- The master branch has been updated by Alan Modra <[email protected]>: https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=9621e7e98df5895eeb2f4aadbdc95687dbf6b8f5 commit 9621e7e98df5895eeb2f4aadbdc95687dbf6b8f5 Author: Alan Modra <[email protected]> Date: Thu Oct 29 08:26:24 2026 +1030 Ignore empty symtab sections Confusion results if we accept an empty .symtab as elf_symtab_hdr. A lot of BFD code for ELF targets supported by GNU ld requires that the STN_UNDEF symbol be treated as a local sym when processing relocs, which means it must be present and sh_info for .symtab be non-zero. This patch stops BFD from attempting to handle relocations in an object with an empty .symtab, as is already the case for relocations where .symtab is missing entirely. More reporting is done too. The extra reporting showed failures on rx-elf when attempting to assemble strip-15rela.s from the binutils testsuite. The test would need to be modified to create .relaP rather than .rela.text to work on that target. I opted instead to disable the test. PR 34835 PR 13962 PR 7023 bfd/ * elf.c (bfd_section_from_shdr): Do not use an empty SHT_SYMTAB or SHT_DYNSYM section. Validate sh_info of SHT_SYMTAB and SHT_DYNSYM sections. Delete bogus SHT_SYMTAB comment. Report an error on bad sh_link or sh_info of reloc sections. binutils/ * testsuite/binutils-all/strip-15.d: Disable for rx. * testsuite/binutils-all/strip-section-headers-1.d: Likewise. * testsuite/binutils-all/strip-13.d: Correct rx comment. * testsuite/binutils-all/strip-14.d: Likewise -- You are receiving this mail because: You are on the CC list for the bug.
